Molecular investigation of oil–water separation using PVDF polymer by molecular dynamic simulation

Abstract

In this study, adsorption of water nanodroplets, oil nanodroplets and oil–water mixtures on a poly vinylidene fluoride (PVDF) surface is investigated computationally by a molecular dynamics (MD) approach and a mechanism for adsorption of the droplets is proposed. MD simulation results revealed that the PVDF surface is hydrophobic and oleophilic and can be used to separate oil from water. The acquired results are in good agreement with experimental results. As to investigate the origins for hydrophobicity of the PVDF surface, mobility of water molecules on the PVDF surface has been meticulously investigated. Investigations have proven that once the water molecules are organized singly or in small groups with two or three molecules, they exhibit very low levels of mobility, whereas water molecules in groups of four or more have high mobility. Some descriptive studies were performed based on MD simulation results to investigate the behavior observed in the systems including: (i) distribution of partial charges on the surface of PVDF. (ii) Dependence of the orientation of water molecules near the PVDF surface on the size of water cluster. (iii) Dependence of the distance of water molecules to the PVDF surface on the water cluster size. Manner of charge distribution on the PVDF demonstrated a formation of nano-domains with positive and negative charges on the surface. Dependence of the behavior of water molecules on the nano-domains on the water cluster size was investigated.

---

1. Introduction

Waste water containing a mixture of emulsified oil–water generated in numerous industrial processes causes damage to the environment and people's health in many cases. Separation of the emulsion mixtures oil–water is usually costly and is a global challenge. One of the ways of water–oil filtration is using the superhydrophobicity and superoleophilicity of porous materials that are permeable to oil or not permeable to water. It is similar to the lotus effect.^1,2 The pressure-driven filtration membranes such as ultrafiltration have been successfully applied for the separation of various emulsions. In addition to being energy consuming, the most serious limitation of filtration membranes is the low flux and quick decline of permeation due to surfactant adsorption and/or pore plugging by oil droplets, which lead to severe fouling issue.³ Materials with superhydrophobic–superoleophilic or superhydrophilic–superoleophobic properties are useful to realize oil–water separation by either filtration or absorption of oils or water from mixtures selectively.^4–7 Jiang et al. reported a superhydrophobic and superoleophilic coating mesh film for the separation of oil and water.⁸ Following this work, a superhydrophilic and underwater superoleophobic hydrogel-coated mesh was recently developed.⁹ PVDF polymer is an advanced material for membrane applications due to its good processability, excellent mechanical properties, and chemical stability. The PVDF polymer has some industrial uses^10,11 for instance in surface cleaning.^12–17 Recently, Xiao et al.¹⁸ developed a PVDF membrane with superhydrophilic and underwater superoleophobic properties for oil/water emulsion separation. Morao et al.¹⁹ fabricated an ultrathin fibrous PVDF membrane with both superhydrophobic and superoleophilic properties. Their investigation showed that fibrous PVDF membranes can be used as high-efficiency liquid separation membranes for separating emulsified water-in-oil solutions.

Zhang et al. reported a methodology to fabricate a super-hydrophobic and super-oleophilic PVDF membrane via an inert solvent-induced phase-inversion process. This membrane can effectively separate both micrometer and nanometer-size surfactant-free and surfactant stabilized water-in-oil emulsions solely driven by gravity, with high separation efficiency.²⁰ Other applications of the PVDF polymer have been reported in micro filtration,²¹ ultrafiltration²² and bioreactors.^23,24

Considering the importance of the water–oil separation, there are numerous experimental studies about the super-hydrophobic and super-oleophilic surface of PVDF;^18–20 however, there are inadequate molecular investigations focusing on the wetting behavior of PVDF. In this work, we study the adsorption behavior of the water droplet, oil droplet and oil–water mixture on the PVDF surface using molecular dynamic simulations. In addition, we investigate the hydrophobicity behavior of the PVDF surface and also the mobility of water clusters on the PVDF surface.

2. Simulation details and methods

To study the adsorption behavior of the water droplet, oil droplet and a mixture of water and oil on the PVDF, a series of MD simulations was performed using the LAMMPS package²⁵ and structures were visualized using the VMD²⁶ package. The multi-component oil droplet in our study was modeled as a ternary mixture of hydrocarbons: heptane, decane and toluene with molecular relative percentage of 60%, 20% and 20%, respectively.²⁷ Non-periodic boundary conditions were applied in the z-direction of the simulation box, whereas periodic boundary conditions were considered in the other two directions.²⁸ The substrate atoms were fixed in their respective initial positions and represent an inert wall.²⁹ The Nosé–Hoover thermostat³⁰ was applied to maintain the temperature at 300 K with damping coefficients 0.1 ps⁻¹. Non-bonded van der Waals interactions were modeled in terms of 12-6 Lennard-Jones famous potentials U_LJ(r_ij)³¹ with a cutoff distance of 10 Å. For the water molecules, the SPC/E potential model was used.³² The OPLS force field was used to model the bond, angle, dihedral, van der Waals and electrostatic interactions between the hydrocarbon molecules.³³ The parameters of the atoms in the PVDF polymer were taken from the work of Lachet et al.³⁴ Tables 1 and 2 list the parameters used for the simulations. For the PVDF polymer, intermolecular interactions and intramolecular interactions are described through a 12-6 Lennard-Jones (LJ) potential:where r_ij, ε_ij, and σ_ij are the interparticle distance, the LJ well depth, and the LJ size, respectively.where q_i is the magnitude of charge i, ε₀ the vacuum permittivity, and r_ij is the distance between charges i and j. The bonded interactions are given by

U_bond(r)/k_B = k_bond(r − r₀)² (3)

where k_B and k_bond are the Boltzmann constant and bonding constant, respectively, and r is the distance between the two bonded atoms, r₀ is the equilibrium bonded distance, and

U_bend(θ)/k_B = k_bend(θ − θ₀)² (4)

where k_bend is the bending constant, θ is the bending angle associated with atoms separated by two bonds, and θ₀ is the equilibrium bending angle;where the χ angle is defined such that χ = ϕ + π, where ϕ and a_n are the dihedral angles associated with atoms separated by three bonds and the corresponding torsional coefficients, respectively.

Table 1 Non-bonded force field parameters for water and PVDF

Atom	εi (kcal mol^−1)	σi (Å)	q (e)	m (au)	Ref. no.
OW	0.15540	3.16600	−0.84760	15.99900	32
HW	0.00000	0.00000	+0.42380	1.008000	32
CPH	0.06600	3.50000	−0.650250	12.01100	34
CPF	0.06600	3.50000	+0.765000	12.01100	34
HP	0.03000	2.50000	+0.225875	1.008000	34
FP	0.06000	2.98300	−0.283250	18.99840	34

---

Table 2 Bonded force field parameters for the PVDF polymer³⁴

Bonding	r0 (Å)	kbond (K Å^−2)
C–C	1.534	310 889
C–H	1.085	329 709
C–F	1.357	502 514

---

Bending	θ0 (deg)	kbend (K rad^−2)
C–C–F	107.74	90 579
F–C–F	105.27	120 772
C–C–C	118.24	80 817
C–C–H	108.45	43 176
C–C–F	109.27	38 748

---

Torsion	an (K)
C–C–C–C	a0 = 920.89
	a1 = 1841.78
	a2 = −1323.47
	a3 = −5042.25
	a4 = 221.42
	a5 = 3421.89
	a6 = 402.57
	a7 = 0.00
	a8 = 0.00
F–C–C–C	a0 = 1069.96
	a1 = 1035.05
	a2 = −1764.99
	a3 = −3636.48
	a4 = 1676.33
	a5 = 2943.89
	a6 = −638.83
	a7 = 0.00
	a8 = 0.00

---

The initial condition of the system: basic configuration of the system is composed of a flat surface of PVDF and a cubic box of the fluids (water, oil or water/oil mixture) located over the surface. In the simulation box, the numbers of water and oil are 1000 and 140 molecules, respectively. The size of the simulation box is set as 100 × 124 × 124 ų in the x, y and z-direction, respectively. Fig. 1 shows that initial configuration of the system used in the simulation. Our MD simulations are conducted in the constant-volume and constant-temperature (300 K) ensemble. The MD time step is set at 0.5 fs. Each MD simulation is 2.0 ns.

Fig. 1 Initial configuration of (a) water–PVDF system and (b) oil–PVDF system. (c-I) Configuration of the monomer unit used in our work. Cyan, white and pink colors represent carbon atom, hydrogen atoms and fluorine atoms, respectively. (c-II) Two hundred molecules of the PVDF in a box. (c-III) Flat PVDF sheet (side view). (c-IV) Flat PVDF sheet (top view).

Construction of the PVDF substrate: typically, for creating a flat polymer, the polymer chains are placed side by side the substrate in the simulation box.^35–40 This method is suitable for linear polymers while the PVDF polymer is a non-linear polymer because of the different sizes of fluorine and hydrogen atoms of the polymer as well as different interactions between the polymer atoms. To create a flat substrate PVDF, we placed 200 polymer chains in the simulation box so that each polymer chain has 5 (CH₂CF₂)₅ monomeric units.⁴¹ It should be noted that the simulation box should be large enough so that the polymer chains be able to move easily. Then, gradually the upper wall of the simulation box gets closer to the lower wall of the simulation box along the z-direction, so that the distance between two walls reaches a distance of 5 angstrom. Then, gradually, the upper wall of the simulation box nears the lower simulation wall of the simulation box along the z-direction so that the distance between two mentioned walls reaches a distance of 5 angstroms. This process was conducted during 5 ns in the NVT ensemble. The time step was 0.5 fs. It should be noted the box dimensions must be large enough in the x and y axes so that by approaching the upper wall to the lower wall along the z axis, the polymer chains can be easily spread in the x–y plane. In order to optimize the polymer, the system was subjected to annealing cycles of 2 ns duration wherein the temperature was increased linearly from 300 K to 800 K. After annealing, the system was cooled to 300 K in 2 ns time intervals.

For the water–oil mixture on the PVDF substrate, the simulations were performed in the NVT ensemble for 10 × 10⁶ time steps wherein each time step was 0.5 fs. Periodic boundary conditions were considered in three directions.

3. Results and discussions

3.1. Behavior of the water droplet, oil droplet and water–oil mixture on the PVDF surface

The determination of the contact angle can be made to investigate fluid-surface interactions. Wettability studies can provide appropriate knowledge about the behavior of the fluid on the surface. Therefore, a first calculation of contact angle can be made to examine the interactions of water and oil with the surface of PVDF. Fig. 2 (top) presents the final configurations of the water–PVDF and oil–PVDF systems. Moreover, to evaluate the stability of the systems during the simulations, the root mean square deviation (RMSD) of the oxygen atoms of water molecules and carbon atoms of the oil in the water–PVDF and oil–PVDF systems were calculated. Fig. 2 (bottom) presents the RMSD diagram of water–PVDF and oil–PVDF systems. The RMSD diagram reveals that water–PVDF and oil–PVDF systems reached a stable state. The acquired fluid-surface contact angles are 4° and 132° for systems of oil–PVDF and water–PVDF, respectively. It is worth mentioning that Yuan et al. reported the water contact angle on the PVDF membrane to be 140°.²² Zhang et al. demonstrated contact angles of 158° and less than 1° for systems of water–PVDF and oil–PVDF, respectively.²⁴ In addition, Yu et al. obtained a water contact angle on the PVDF polymer of 130°.⁴² Moreover, the water contact angle of 153° and oil contact angle of 0° has been reported for ultrathin fibrous PVDF membranes.¹⁹ It is important to emphasize that regarding the synthesis method of polymeric membranes, polymers have various levels of compression, which has an impact on the contact angle of the water and oil on the polymer surface. The obtained contact angle values for describing systems reveal that the PVDF surface is a hydrophobic and oleophilic surface. Accordingly, it can be employed for separation of oil–water systems. As to deeply investigate the level of oil–water separation, another simulation was conducted.

Fig. 2 Final configuration and RMSD diagram of water–PVDF system (right) and oil–PVDF (left).

Fig. 3 presents the initial and final configurations of the abovementioned systems after 5 ns simulation time and also the RMSD diagram for oil–water–PVDF. The RMSD diagram shows that the system reached a stable state. Fig. 3 compares the oil molecules adsorbed to the PVDF surface with the water molecules not adsorbed by the surface. It is also worth mentioning that water molecules forming droplets are separated from the surface, but small molecules aggregated on the surface of the PVDF are also observed. The relative percentages of oil and water molecules in the z-direction are shown in Fig. 4. From Fig. 4, it is apparent that oil molecules are assembled close to the PVDF surface, while water molecules aggregation is formed at a further distance to the surface. On balance, the whole of these observations examine the superiority of the PVDF surface in separation of water from oil.

Fig. 3 (a) Initial configuration (b) final configuration (c) RMSD diagram of the water–oil–PVDF system.

Fig. 4 Relative percentages of oil and water molecules in the z-direction, O is oxygen atom of water and C refers to carbon atoms of heptane, decane and toluene molecules.

3.2. Adsorption and orientation of oil and water molecules on the surface of PVDF

As argued in the previous section, the PVDF polymer is oleophilic and hydrophobic surface. As to probe the underlying mechanism behind hydrophobicity and oleophilicity of the PVDF surface, the adsorption mechanism of water and oil molecules on the surface is discovered. Insights into the adsorption mechanism can provide the information on how molecules approach the surface and their effective interactions. The PVDF polymer consisted of four types of atoms, including F, H, C connected to F and C connected to H, which have charges of −0.283250 e, 0.225875 e, 0.765000 e and −0.650250 e, respectively. Possession of these various charges in the polymer leads to different electrostatic interactions between the polymer surface and the fluids placed on it. Furthermore, the van der Waals interactions between each pair of polymer and fluid atoms are distinct. When the electrostatic and van der Waals interactions of PVDF with oil–water fluid is taken into consideration, it is possible to study the adsorption mechanism of oil and water molecules on the surface. Table 3 shows the van der Waals interaction between the oxygen atoms of water and atoms of the PVDF polymer that are located 4 Å apart. Due to the force field used for water molecules, the van der Waals interaction between the water hydrogen atom and atoms of the PVDF polymer is zero. The van der Waals interactions between the oxygen atom of water and fluorine atom of the polymer cause the water molecules to get close to the PVDF surface. The electrostatic repulsion force between the oxygen atoms of water and fluorine atoms of the polymer and also the attraction force between the oxygen atoms of water molecules and hydrogen atoms of the PVDF exerted force on the water molecule. Water molecules subjected to these forces move toward the hydrogen atoms on the PVDF surface via their oxygen atoms.

Table 3 The van der Waals interaction energy between the oxygen atoms of water and the PVDF atoms

	ε (kcal mol^−1)	σ (Å)	van der Waals interaction (kcal mol^−1)
OW–CPF	0.10127	3.333	−0.09020116
OW–CPH	0.10127	3.333	−0.09020116
OW–HP	0.06827	2.833	−0.03011707

---

The results obtained from the radial distribution function (RDF) analysis are presented in Fig. 5 for the hydrogen atom in polymer and oxygen atom in water. In Fig. 5a, the RDF diagram for polymer hydrogen–water oxygen reveals a sharp peak at a distance of 2.4 Å and a value of 7.1. On the other hand, the RDF diagram for polymer hydrogen–water hydrogen shows a sharp peak at a distance of 3.38 Å and a value of 3. These values reveal that the oxygen atom of water is placed closer to the surface compared to hydrogen atoms of water. In other words, this oxygen atom possesses strong interaction with hydrogen atoms on the surface. It is apparent from Fig. 5b that hydrogen atoms of PVDF are the sites of adsorption satisfied by the oxygen atoms of water. Similar to water molecules, toluene molecules are intended to move toward PVDF. Orientation of toluene molecule occurs by the parallel placement of the benzene group relative to PVDF due to highest interaction between toluene and PVDF. At closer distances of toluene–PVDF, electrostatic interactions dominate. The negative charge is centralized on the benzene group, whereas the positive charges are distributed on the hydrogen atoms. The orientation of toluene molecule close to PVDF occurs such that the benzene group and hydrogen atoms of toluene possess close distance to positive and negative charged segments of the surface. In Fig. 6, the variation of the angle between benzene planes in toluene and the plane of polymer surface is presented. Moreover, a schematic showing the orientation of the toluene molecule near the polymer is revealed. As shown in Fig. 6a, the average angle of the benzene group plane and polymer surface is less than 20°, which explains the toluene tendency approaching the PVDF hydrogen atoms by its benzene functional group. A contributory factor in different adsorption water and toluene molecules on the polymer surface is size discrepancy fluid molecules. Due to a lower kinetic diameter in water molecules, the presence of water molecules in the nano-domains is likely to occur. Consequently, strong interactions of the water and PVDF atoms in nano-domains require water molecules to possess low levels of mobility. However, large toluene molecules cannot completely fit in the mentioned space and are therefore forced to float on the surface of PVDF. As with the mechanisms proposed for adsorption of water and oil molecules, water molecules seem to have the highest PVDF adsorption, whereas contact angle measurements for water droplet prove weak water–PVDF interactions. The justifications for this statement are included in the following sections.

Fig. 5 (a) RDF diagram for water hydrogen–polymer hydrogen (black line) and water oxygen–polymer hydrogen (red line). H_p, H_w and O_w are hydrogen atom of polymer, hydrogen and oxygen atoms of water, respectively. (b) A schematic of water molecule adjacent to the PVDF surface. Water molecules approach the PVDF surface through their oxygen atom to acquire minimum distance from polymer hydrogen. Blue, white, pink and red colors denote the carbon, hydrogen, fluorine and oxygen atoms.

Fig. 6 (a) Variation of angle between the benzene plane in toluene and the plane of polymer surface. (b) Schematic of benzene adjacent to the PVDF surface. Toluene molecules approach the PVDF surface by the benzene group to have a minimum distance from hydrogen atoms placed on the surface. Blue, white and pink colors denote the carbon, hydrogen, fluorine and oxygen atoms.

3.3. Investigation of water clusters distribution and mobility on the PVDF surface

In order to justify the proper interaction between single water molecules on the PVDF surface, while the PVDF polymer is a hydrophobic surface, diagrams of water motilities on the PVDF surface with respect to cluster size have been acquired. For this purpose, 10 simulations with variations of only the size of water clusters on the polymer surface were conducted. Fig. 7 shows snapshots of the water–polymer system at different times for one water molecule and a cluster with four water molecules. Fig. 7 shows that one water molecule has low mobility; however, clusters of 4 molecules move along the PVDF surface.

Fig. 7 Snapshots of water–polymer system at different times for one water molecule and a cluster with four water molecules.

Fig. 8 shows the movement of different clusters on the PVDF surface through the last 0.75 ns of simulations. Fig. 8 further shows that clusters with one, two or three water molecules have low mobility; however, clusters of 4 or more molecules move along the PVDF surface. Such a behavior is considered to be a special case that was observed in the PVDF surface through MD simulations. Three determinate factors are considered to justify the abovementioned behavior:

Fig. 8 Movement of different clusters on the PVDF surface during the last 0.75 ns of simulations.

(i) Distribution of partial charges on the PVDF surface

(ii) Dependence of the water molecules orientation close to the PVDF surface on the size of the water cluster

(iii) Dependence of distance of water cluster from the PVDF surface on the size of the water cluster

As previously stated, the electrostatic interactions play a critical role on how fluid molecules adsorb onto the PVDF surface. A diagram of partial charge distribution on the PVDF surface was plotted. In order to plot partial charge distribution, a point in the XY plane with +1 formal charge and 3 angstrom was scanned and the respective electrostatic interaction energies were calculated with all atoms present in the surface. Fig. 9 depicts the distribution of electrostatic energy on the PVDF surface. Red and blue regions show accumulation of positive and negative charges on the PVDF surface, respectively. Green and yellow regions correspond to regions with almost homogeneous charge distribution.

Fig. 9 Electrical energy distribution diagram on the PVDF surface. Sections in red and blue colors show regions with the highest positive and negative charges, respectively.

From Fig. 9, it can be concluded that PVDF has regions with accumulation of charges with the same sign. As with the mechanism proposed for the adsorption of water molecules on the PVDF surface, water molecules tend to be positioned in regions with accumulation of positive charges to have maximum interactions with the surface. Fig. 9 exhibits charge accumulation regions to be discretely dispersed. Since water molecules do not tend to have distances from positively charged regions, they are then discretely dispersed on the polymer surface. A decrease in the mobility of water molecules in small groups with one, two or three water molecules can then be attributed to the special charge distribution on the PVDF surface. Furthermore, in the polymer surface, the number of nano-domains with positive charges is more than the number of nano-domains with negative charges. The dominance of positively charged regions over negatively charged regions in a neutral primer can be attributed to the smaller size of hydrogen atoms in the polymer. Small sizes of hydrogen atoms cause the atoms to be able to get close to each other and to be more compactly organized relative to fluorine atoms in the polymer. Such a dense organization will lead to the accumulation of positive charges on the PVDF surface. The RDF diagrams for polymer-hydrogen and polymer-hydrogen as well as polymer-fluorine and polymer-hydrogen were plotted as proof.

Fig. 10 shows the polymer hydrogen–polymer hydrogen RDF as well as polymer fluorine–polymer fluorine RDF. The polymer hydrogen–polymer hydrogen RDF exhibits a sharp peak of 1.72 Å distance and a value of 21.6, whereas a sharp peak at 2.2 Å and a value of 15.9 is observed for the polymer fluorine–polymer fluorine RDF. As it can be seen from Fig. 10, hydrogen atoms are closer than fluorine atoms in the PVDF polymer. Further investigation of cluster orientation and spatial direction of dipole of water molecules on the polymer surface is carried out. Fig. 11 represents the average spatial orientation of the water dipole moment with respect to the PVDF surface for the last 0.5 ns simulation time. In Fig. 11, it is seen that low number of water molecules in clusters leads to significant interactions of water and polymer molecules and causes the water polymer relative angle to be more than 90°. As with the increase in cluster sizes, slipping of clusters on the surface occurs more easily. It clarifies the less mobility of small clusters compared to clusters with a significant number of water molecules. It is worth mentioning that smaller clusters tend to place in positively charged nano-domains. Another significant factor is the expansion of cluster volumes by the increase of water molecule numbers. With the increase of cluster size, water molecules can barely diffuse into the nano-domain regions. Lower diffusion of water leads to the placement of water molecules at further distances with respect to the surface alongside the weakening of water–polymer surface interactions. In Fig. 12, the minimum distance of oxygen atoms in water from the PVDF surface for various cluster sizes is shown for the last 0.5 ns simulation time. Results obtained for Fig. 11 and 12 are shown schematically in Fig. 13. Fig. 13 shows the interaction of a water molecule and a few water molecules with the PVDF substrate. Fig. 13 shows that by increasing the number of water molecules from one to four times, the water–water interaction overcomes water–polymer interactions, which alters the orientation of water molecules. This orientation change weakens water–polymer interactions; therefore, water molecules are released from the surface and move to other surfaces.

Fig. 10 Radial distribution functions, H_p–F_p respectively, for polymer hydrogen and polymer fluorine.

Fig. 11 Average spatial orientation of water dipole moments with respect to the PVDF surface for the last 0.5 ns simulation time.

Fig. 12 Average distance of the mass center of water cluster from the surface of the polymer. Smaller clusters are closer to the surface in comparison with larger clusters.

Fig. 13 Interaction of a molecule of water (a) and water cluster (b) with the PVDF substrate.

4. Conclusion

In the present study, the wettability of the PVDF polymer by water and oil molecules and also the hydrophobic properties of PVDF polymer were studied. The proposed adsorption mechanism of the water and oil molecules on the PVDF surface reveals that the presence of the positively and negatively charged nano-domains on the PVDF surface affects the positioning and orientation of the water molecules. Each nano-domain included one, two or three water molecules, whereas water clusters with four molecules or more accumulated on the PVDF surface and the interaction between water the PVDF surface was reduced. As with the increase in the water cluster size, the interactions of water molecules are intensified along with the weakening of water–PVDF interactions. Due to a large volume of water clusters, they cannot permeate the nano-domains on the PVDF surface and appropriate interactions between them and the surface are not made. The obtained results showed that water molecules in the form of single, binary or triple exhibit very low levels of mobility, whereas water molecules in groups of four or more have high mobility.

References

1. C. Neinhuis and W. Barthlott, Ann. Bot., 1997, 79, 667–677.
2. W. Barthlott and C. Neinhuis, Planta, 1997, 202, 1–8.
3. M. M. Pendergast and E. M. Hoek, Energy Environ. Sci., 2011, 4, 1946–1971.
4. J. Yuan, X. Liu, O. Akbulut, J. Hu, S. L. Suib, J. Kong and F. Stellacci, Nat. Nanotechnol., 2008, 3, 332–336.
5. A. Li, H. X. Sun, D. Z. Tan, W. J. Fan, S. H. Wen, X. J. Qing and W. Q. Deng, Energy Environ. Sci., 2011, 4, 2062–2065.
6. Y. Zhang, S. Wei, F. Liu, Y. Du, S. Liu, Y. Ji and F. S. Xiao, Nano Today, 2009, 4, 135–142.
7. C. H. Lee, N. Johnson, J. Drelich and Y. K. Yap, Carbon, 2011, 49, 669–676.
8. L. Feng, Z. Zhang, Z. Mai, Y. Ma, B. Liu, L. Jiang and D. Zhu, Angew. Chem., Int. Ed. Engl., 2004, 43, 2012–2014.
9. Z. Xue, S. Wang, L. Lin, L. Chen, M. Liu, L. Feng and L. Jiang, Adv. Mater., 2011, 23, 4270–4273.
10. N. Yoshida, M. Takeuchi, T. Okura, H. Monma, M. Wakamura, H. Ohsaki and T. Watanabe, Thin Solid Films, 2006, 502, 108–111.
11. Y. Li, W. Cai, G. Duan, B. Cao, F. Sun and F. Lu, J. Colloid Interface Sci., 2005, 287, 634–639.
12. S. H. Park, S. M. Lee, H. S. Lim, J. T. Han, D. R. Lee, H. S. Shin and J. H. Cho, ACS Appl. Mater. Interfaces, 2010, 2, 658–662.
13. P. Muthiah, S. H. Hsu and W. Sigmund, Langmuir, 2010, 26, 12483–12487.
14. Y. Chen and H. Kim, Appl. Surf. Sci., 2009, 255, 7073–7077.
15. Z. Zheng, Z. Gu, R. Huo and Z. Luo, Appl. Surf. Sci., 2010, 256, 2061–2065.
16. B. B. J. Basu and A. K. Paranthaman, Appl. Surf. Sci., 2009, 255, 4479–4483.
17. E. Bormashenko, T. Stein, G. Whyman, Y. Bormashenko and R. Pogreb, Langmuir, 2006, 22, 9982–9985.
18. Y. Xiao, X. D. Liu, D. X. Wang, Y. K. Lin, Y. P. Han and X. L. Wang, Desalination, 2013, 311, 16–23.
19. A. Morao, A. B. Alves and J. P. Cardoso, Sep. Purif. Technol., 2001, 22, 459–466.
20. Z. Badani, H. Ait-Amar, A. Si-Salah, M. Brik and W. Fuchs, Desalination, 2005, 185, 411–417.
21. M. Brik, P. Schoeberl, B. Chamam, R. Braun and W. Fuchs, Process Biochem., 2006, 41, 1751–1757.
22. T. Yuan, J. Meng, T. Hao, Z. Wang and Y. Zhang, ACS Appl. Mater. Interfaces, 2015, 7, 14896–14904.
23. Z. Zhou and X. F. Wu, Mater. Lett., 2015, 160, 423–427.
24. W. Zhang, Z. Shi, F. Zhang, X. Liu, J. Jin and L. Jiang, Adv. Mater., 2013, 25, 2071–2076.
25. S. Plimpton, J. Comput. Phys., 1995, 117, 1–19.
26. W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics, 1996, 14, 33–38.
27. F. C. Wang and H. A. Wu, Soft Matter, 2013, 9, 7974–7980.
28. Z. Xue, S. Wang, L. Lin, L. Chen, M. Liu, L. Feng and L. Jiang, Adv. Mater., 2011, 23, 4270–4273.
29. M. M. Pendergast and E. M. Hoek, Energy Environ. Sci., 2011, 4, 1946–1971.
30. W. G. Hoover, Phys. Rev. A, 1985, 31, 1695–1697.
31. J. E. Lennard-Jones, Proc. Phys. Soc., 1931, 43, 461–482.
32. H. J. C. Berendsen, J. R. Grigera and T. P. Straatsma, J. Phys. Chem., 1987, 91, 6269–6271.
33. W. L. Jorgensen, D. S. Maxwell and J. Tirado-Rives, J. Am. Chem. Soc., 1996, 118, 11225.
34. V. Lachet, J. M. Teuler and B. Rousseau, J. Phys. Chem. A, 2014, 119, 140–151.
35. S. Chen, J. Wang, T. Ma and D. Chen, J. Chem. Phys., 2014, 140, 114704–114711.
36. J. T. Hirvi and T. A. Pakkanen, J. Chem. Phys., 2006, 125, 144712–144722.
37. J. T. Hirvi and T. A. Pakkanen, J. Chem. Phys., 2007, 111, 3336–3341.
38. J. T. Hirvi and T. A. Pakkanen, Surf. Sci., 2008, 602, 1810–1818.
39. Y. Jiang, J. T. Hirvi, M. Suvanto and T. A. Pakkanen, Chem. Phys., 2014, 429, 44–50.
40. D. R. Nieto, F. Santese, R. Toth, P. Posocco, S. Pricl and M. Fermeglia, ACS Appl. Mater. Interfaces, 2012, 4, 2855–2859.
41. G. Zhu, Z. Zeng, L. Zhang and X. Yan, Comput. Mater. Sci., 2008, 44, 224–229.
42. Y. Yu, H. Chen, Y. Liu, V. S. Craig, C. Wang, L. H. Li and Y. Chen, Adv. Mater. Interfaces, 2015, 2, 1400267.

---